Jetzt zum Sommersemester an der TU Hamburg bewerben! Bewerbungen für deutschsprachige und ausgewählte englischsprachige Masterprogramme sind bis zum 15. Januar möglich.

Microharvest

Project result: Poster (PDF)

Pi Eye - real-time vision detection system

Description of the company

MicroHarvest is a fast-growing startup based in Hamburg, Germany, specializing in the production of sustainable proteins from microorganisms. Through the development of these environmentally friendly protein sources, MicroHarvest is positioning itself as a pioneer in addressing the world's food challenges by providing a sustainable and innovative solution that actively contributes to improving global food security.

Problem

Monitoring of small bioreactors requires quick action if and when issues arise. Especially in the case of over-night fermentations, it is not possible to have researchers present in the lab 24/7. The acceleration of AI technology combined with the decreasing price of the necessary hardware means that these technologies can be better taken advantage of by small companies.

Aims of the project
The project's primary aim is to implement a real-time vision detection system using a Raspberry Pi or Arduino based system. This system will be capable of visually monitoring a fermentation process to identify anomalies, ensuring operational safety and efficiency. The project will involve developing an AI-based image recognition system that can detect deviations from expected procedures and alert the laboratory team to potential issues.

Scopes

1.Literature Review and Technology Selection
o Research and evaluate the latest open-source AI vision models suitable for real-time processing on devices like the Raspberry Pi.

2 Data Collection and Labeling
o Extract frames from archived video sources, label them accurately to create a dataset for training the AI model.

3. Model Training and Evaluation
o Implement the AI model using machine learning libraries, train it on the labeled dataset, and evaluate its performance.

4. Real-Time Detection System Development
o Develop a system integrating the trained model for live video processing and alert generation in case of anomalies.

Target group (students)

  • Data Science: Students specializing in data analysis, machine learning, and AI model development. Interested in applying statistical analysis and algorithmic processing to derive insights and implement real time detection.
  • Computer Science: Students with a strong foundation in programming, software development, and system integration. Interested in developing applications, including AI and machine learning projects.
  • Electrical Engineering: Students focused on electronic systems, signal processing, and hardware-software integration. Interested in the practical application of electronics in computing systems, including the deployment of AI models on devices like the Raspberry Pi.

Dates
Please save these dates: Fishing for Experience Termine

Registration
You can apply for Fishing for Experience online.